Model Transformation by Partial Evaluation of Model Interpreters

In model-driven development, the use of both model translators and model interpreters is widespread. It is also well-known that partial evaluation can turn an interpreter into a translator. In this paper the authors show that a simple online partial evaluator is effective at specializing a model interpreter with respect to a model to create a compiled model interpretation. Data models pose a particular problem, because it is not clear what a data model interpreter would do, given that data is generally considered to be passive. They show how a data model interpreter can be defined in an object-oriented style as a dynamic message-processing function.